flywp / includes / StaticSite / Engine / Discovery / UrlDiscovery.php

UrlDiscovery.php in FlyWP Helper – Page Cache, Page Optimization, Emails for FlyWP Server Control Panel 1.8.0, at includes/StaticSite/Engine/Discovery/UrlDiscovery.php

155 lines 4.3 KB

Raw Download Zip

1	<?php
2
3	namespace FlyWP\StaticSite\Engine\Discovery;
4
5	/**
6	* Collects the start URLs of a full build from all sources, and the URLs that must never be published.
7	*
8	* - Multisite is out of scope. The engine builds the site of the current home URL.
9	*
10	* @since 1.8.0
11	*/
12	class UrlDiscovery {
13
14	/**
15	* Rows in one query.
16	*/
17	const BATCH_SIZE = 500;
18
19	/**
20	* @var WordPressGateway
21	*/
22	private $wp;
23
24	/**
25	* @var string\|null
26	*/
27	private $origin_url;
28
29	/**
30	* @var UrlSource[]
31	*/
32	private $sources;
33
34	/**
35	* @param WordPressGateway $wp Gateway.
36	* @param string\|null $origin_url Origin URL that the engine fetches. Null to use the home URL.
37	* @param UrlSource[] $sources Sources, in order.
38	*/
39	public function __construct( WordPressGateway $wp, $origin_url, array $sources ) {
40	$this->wp = $wp;
41	$this->origin_url = $origin_url;
42	$this->sources = $sources;
43	}
44
45	/**
46	* @param WordPressGateway $wp Gateway.
47	* @param string\|null $origin_url Origin URL.
48	*
49	* @return UrlSource[]
50	*/
51	public static function default_sources( WordPressGateway $wp, $origin_url ) {
52	return [
53	new HomeSource( $wp ),
54	new PostTypeSource( $wp ),
55	new TaxonomySource( $wp ),
56	new AuthorSource( $wp ),
57	new DateArchiveSource( $wp ),
58	new FeedSource( $wp ),
59	new SitemapSource( $wp ),
60	new SiteFileSource( $wp ),
61	new FilterSource( $wp, $origin_url ),
62	];
63	}
64
65	/**
66	* URLs on the origin, unique by URL. URLs on other hosts are removed.
67	*
68	* @return \Generator<DiscoveredUrl>
69	*/
70	public function urls() {
71	$origin = new OriginUrl( $this->wp->home_url(), $this->origin_url );
72	$seen = [];
73
74	foreach ( $this->sources as $source ) {
75	foreach ( $source->urls() as $item ) {
76	$url = $origin->rebase( $item->url );
77
78	if ( null === $url ) {
79	continue;
80	}
81
82	$key = md5( $url, true );
83
84	if ( isset( $seen[ $key ] ) ) {
85	continue;
86	}
87
88	$seen[ $key ] = true;
89	$item->url = $url;
90
91	yield $item;
92	}
93	}
94	}
95
96	/**
97	* URLs that must never be published, on the origin: password-protected published posts, and posts and terms
98	* that Yoast SEO or Rank Math marks as noindex. Queries run in batches.
99	*
100	* @return \Generator<string>
101	*/
102	public function excluded_urls() {
103	$origin = new OriginUrl( $this->wp->home_url(), $this->origin_url );
104
105	foreach ( $this->excluded_links() as $link ) {
106	$url = $origin->rebase( $link );
107
108	if ( null !== $url ) {
109	yield $url;
110	}
111	}
112	}
113
114	/**
115	* @return \Generator<string\|null>
116	*/
117	private function excluded_links() {
118	$post_types = $this->wp->public_post_types();
119	$after_id = 0;
120
121	do {
122	$rows = $this->wp->post_rows( $post_types, $after_id, self::BATCH_SIZE );
123	$full_batch = count( $rows ) === self::BATCH_SIZE;
124	$noindex = RobotsMeta::noindex_post_ids( $this->wp->seo_robots_meta( array_column( $rows, 'id' ) ) );
125
126	foreach ( $rows as $row ) {
127	$after_id = $row['id'];
128
129	if ( PostTypeSource::is_excluded( $row, $noindex ) ) {
130	yield $this->wp->permalink( $row['id'] );
131	}
132	}
133	} while ( $full_batch );
134
135	foreach ( RobotsMeta::yoast_noindex_terms( $this->wp->seo_term_meta_option() ) as $term ) {
136	yield $this->wp->term_link( $term['id'], $term['taxonomy'] );
137	}
138
139	$after_id = 0;
140
141	do {
142	$rows = $this->wp->seo_term_robots_meta( $after_id, self::BATCH_SIZE );
143	$full_batch = count( $rows ) === self::BATCH_SIZE;
144
145	foreach ( $rows as $row ) {
146	$after_id = (int) $row['term_id'];
147
148	if ( RobotsMeta::is_rank_math_noindex( $row['meta_value'] ) ) {
149	yield $this->wp->term_link( $row['term_id'], $row['taxonomy'] );
150	}
151	}
152	} while ( $full_batch );
153	}
154	}
155